Outside of statistical process control, current manufacturing quality control is founded upon a “make-then-inspect” mindset . While this approach is an important part of the quality control process, post-process inspection is labor intensive, a bottleneck to continuous production throughput, costly, subject to human interpretation, and susceptible to missing subtle defects. This paper presents the application of in-process quality control (IPQC) during inertia friction welding of critical components.
This paper is a follow-on to a preliminary investigation into a new sensing technique for real-time inspection of product quality during friction welding . The previous effort explored the feasibility of modeling the approach that an experienced friction welding operator uses to distinguish anomalous process behavior during friction welding. In particular, a non-contact, audio-based sensor was used to capture the audible process dynamics during inertia friction welding of a dual-alloy component. The previous work employed a neural-network-based data mining technique to locate and identify features within the audio data that can be used to discriminate acceptable from unacceptable process behavior. This paper extends the previous work by providing a formal methodology for automatic, real-time, nondestructive, inspection of rotary friction welding.